Table 1.
MK table.
Table 2.
SnIPRE coefficients and population genetic parameters.
Figure 1.
Example joint distribution of the estimated selection effect and the constraint effect for a particular gene.
Data simulated using PRFREQ. The blue asterisk denotes the true location of parameters.
Table 3.
SnIPRE predicted mutation counts.
Table 4.
Expected mutation counts.
Table 5.
False positive rate.
Table 6.
False positive rate and demography.
Figure 2.
Top: Distribution 1, 2, and 3 of used in the coalescent simulations for Table 7. Bottom: Proportion of constraint effects classified as significant by SnIPRE; x-axis is true proportion of non-lethal mutations,
.
Table 7.
Realized coverage of 95% CI for and γ when
,
varies, and there is linkage among sites.
Figure 3.
Comparison of estimates of constraint when (no constraint).
A: The distribution of constraint estimates. B: Constraint estimates versus the selection strength.
Figure 4.
Classification of selection effect for Drosophila-like simulations.
Shaded regions of histogram represent the proportion of genes under selection classified as under selection; x-axis is true selection coefficient; .
Figure 5.
Classification of selection effect for human-like simulations.
Shaded regions of histogram represent the proportion of genes under selection classified as under selection; x-axis is true selection coefficient; .
Table 8.
Selection classification for simulations by method.
Figure 6.
True positive rate versus false discover rate.
Results for data set of 2,000 genes, 550 of the genes are under selection with or
.
Figure 7.
Distribution of residuals for selection coefficient estimates by method.
The top row displays the distribution of constraint, the middle row displays residuals for simulations using ; the bottom row displays residuals for simulations using
. Residuals grouped by true selection strength.
Figure 8.
D. simulans estimated selection effects and non-synonymous effects for 8,887 genes.
Plots A and B shows the estimated selection effects using SnIPRE and B SnIPRE respectively.
Figure 9.
Human estimated selection effects and non-synonymous effects for 11,624 genes.
Plots A and B shows the estimated selection effects using SnIPRE and B SnIPRE respectively. B SnIPRE classifies far more genes as having a negative average selection effect, and this difference can be explained in part by the construction of 95% confidence interval versus the credible interval.